Yelp Dataset Analysis - part2

Import Libraries

library("jsonlite")
library("ggplot2")
library("readr")
setwd("~/Documents/Project/YelpAnalysis/")

Import Business Dataset

business <- "dataset/business.json"
review <-read_lines(business, n_max = 200000)
business.df <- fromJSON(paste("[", paste(review, collapse = ","), "]"))
business.df <- flatten(business.df)
business.df

Investigate variable correlations with MCMC regression

Encode all the caterogries

# one-hot-encode all the categories
library(qdapTools)
categories <- mtabulate(business.df$categories)
categories

create a count dataframe to count the frequency of each unique category and then sum them, Keep the categories with the top 15 frequency

# create a count dataframe to count the frequency of each unique category and then sum them
categories.sum <- data.frame(
                 category=character(),
                 counts=integer(),
                 stringsAsFactors=FALSE)
for (name in names(categories)) {
  categories.sum[nrow(categories.sum) + 1,] = list(name, sum(categories[name]))
}
categories.sum

# Keep the top 15 categories
eliminated.categories <- categories.sum[order(categories.sum$counts, decreasing = TRUE),]
eliminated.categories <- eliminated.categories[1:15,]
eliminated.categories

# remove all the column that is not the top 15 category
library(dplyr)
categories.filtered <- categories[eliminated.categories$category]

merge the categories back to business dataframe, Only keep the target column, the explanatory columns(categories column) and business id and name

categories.filtered$ID <- seq.int(nrow(categories.filtered))
business.df$ID <- seq.int(nrow(business.df))
encoded.business.df <- left_join(business.df[,c("ID", "business_id", "name", "stars")], categories.filtered)

Joining, by = "ID"

encoded.business.df

Apply MCMC Linear Regression

plot(lm1)

Error in plot(lm1) : object 'lm1' not found